Bayesian QTL Mapping And Genetic Architcture Dissection For Barley Underlining Important Traits

Linkage analysis is one of the important method for QTL detection,which utilizes information from whole genome markers,so that it is able to detect QTL genome-wide.There are some sophisticated methods developed for QTL detection,excellent methods are able to increase QTL detection power and decrease the false-positive rate,and are also able to explore the genetic structure.This study mainly develops new statistical methods and untilizes the current methods for QTL mapping,which aims to increase QTL detection power and decrease false positive rate;and to map interaction QTL for exploring genetic structure.The research includes three parts.Methodologically,two new methods for QTL significance test are developed.Two methods take advantage of two-peak character of QTL posterior distribution and employ the fitting(Fit Mix)and intuitive method(Sim Mix)respectively to separate two-peak distibuiton into two independent distributions,one has mean close to zero and another has mean deviated from zero;then calculate the posterior inclusion probability of a QTL by the proportion of the distribution with non-zero mean.The two methods have good performance on QTL detection in the analysis of American barley and also provide further evidence for other five methods.Two methods Fit Mix and Sim Mix have very similar performance with simiulation and real-data study,and the correlation coefficient reaches 0.81;however,compared with Sim Mix,Fit Mix has lower QTL detection resolution and power especially when the QTL posterior is extreamly low.The performance of two methods are validated with serise of simulations,showing that compared with DE and ST,BIDE and BIST are able to shrink “insignificant” QTL to zero;further,the order of the performance of the four methods is BIDE≈BIST> ST>DE.The research develops two QTL test methods and proposed to detect QTL with high confidence by integrating several methods.It studies on eight quantitative traits of American barley including Alpha amylase,diastatic power,malt extract,protein,grain yield,height,lodging resistance and heading time.For increasing QTL detection power and minimizing the false positive rate,we develop two new QTL-test methods to infer QTL posterior probability and apply two methods together with five current high resolution methods including the single QTL model andfour Bayesian methods: Bayesian LASSO(DE),Student’t prior(ST),improved Bayesian LASSO(BIDE),improved Student’t prior(BIST),to anlyaze these data.Through the interation of 7methods,26 QTL are detected,in which 4 QTL affects Alpha amylase,7 QTL affects diastatic power,2 QTL affects malt extract,4 QTL affects protein,1 QTL affects grain yield,4 QTL affects height,3 QTL affects lodging resistance and 1 QTL affects heading time.Compared with single QTL scan,Bayesian multiple-QTL model totally detected 7 more QTL.The results show that the disadvantage of the single QTLmodel is that it is dificult to distinguish two close linked QTL,and in this situation,two QTL are either combined into one signal or canceled out each other,whereas multiple QTL model has good performance in this way.In additional to the QTL mapping for these traits,this study also investigates the genetic structure of eight traits for American barley by using Bayes B combined with empirical Bayes method.Totally 16 main effect QTL and 8 interacting effect QTL are detected,among which 3main effect QTL and 1 interacting effect QTL are detected for Alpha malt extract;3 main effect QTL are detected for diastatic power;2 main effect QTL and 1 interacting QTL are found underling malt extract;1 main effect QTL controling protein content are detected;2 main effect QTL and 3 interacting QTL controling yield are detected;2 main effect QTL are found for height;1 main effect QTL and 3 interacting effect QTL are found for lodging resistance properties;2 main effect QTL controling flower time are detected.In summary,the proportion of the interacting effect QTL is 33%,explaining large proportion of genetic structure,reflecting the complex genetic mechanism of quantitative trait.